Estimating the Underground Economy using MIMIC Models

Abstract

MIMIC models are being used to estimate the size of the underground economy or the tax gap in various countries. In this paper I examine critically both the method in general and three applications of the method by Giles and Tedds (2002), Bajada and Schneider (2005) and Dell’Anno and Schneider (2003). Connections are shown to familiar econometric models of linear regression and simultaneous equations. I also investigate the auxiliary procedures used in this literature, including differencing as a treatment for unit roots and the calibration of results using other data. The three applications demonstrate how the method is subjective and pliable in practice. I conclude that the MIMIC method is unfit for the purpose.